st: System GMM with xtabond2

[Florian Schwab <florian@le-schwab.ch>]

Hello Statalist

for a panel of individuals, I would like to estimate a function
explaining the probability of being arrested (y) dependent on

- the lagged dependent variable y(i, t-1), to say: having been arrested the
year before
- a vector X(x1(i,t-1), ..., x10(i,t-1)) of the individuals characteristics
(first lag), all of which are used as instruments
- a vector Z(z1, ..., z4) of presumably strictly exogenous variables
such as overall criminality rate, labor market participation rate etc.

my corresponding Stata command is

xtabond2 y L.y x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 z1 z2 z3 z4, gmm(y x1 x2
x3 x4 x5 x6 x7 x8 x9 x10, lags(1 1) collapse) iv (z1 z2 z3 z4) twostep
robust

However, I don't get a significant effect of L.y which is fairly unlikely
since all comparable
studies find one.

Is there something wrong with the specification of the xtabond2 command ?

Thanks in advance.
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/